PaperlayerView exact revision on arXiv

Source-linked AI summary

Recent Advances in the Internet of Medical Things (IoMT) Systems Security

Ali Ghubaish, Tara Salman, Maede Zolanvari, Devrim Unal, Abdulla Al-Ali, Raj Jain

arXiv:2302.04439v1cs.CR

TL;DR

IoMT systems support healthcare monitoring but face major security challenges because sensitive patient data must be protected across collection, transmission, and storage. The paper reviews attacks, requirements, and state-of-the-art techniques, then proposes a combined framework. The framework is designed to cover IoMT security requirements and mitigate the 14 attacks discussed.

  • Problem

    IoMT systems face major security challenges because sensitive healthcare data must be protected across collection, transmission, and storage, while existing techniques do not comprehensively address known attacks.

  • Method

    The paper reviews IoMT security requirements, attack types, and countermeasures, then combines features of multiple techniques into a framework covering collection, transmission, and storage.

  • Results

    The proposed framework is capable of protecting IoMT systems from the 14 discussed attacks and fulfills their required security requirements.

  • Takeaways & Limitations

    Comprehensive IoMT protection requires combining security techniques across all data and device security stages rather than relying on one technique.

Abstract

from arXiv · show

The rapid evolutions in micro-computing, mini-hardware manufacturing, and machine to machine (M2M) communications have enabled novel Internet of Things (IoT) solutions to reshape many networking applications. Healthcare systems are among these applications that have been revolutionized with IoT, introducing an IoT branch known as the Internet of Medical Things (IoMT) systems. IoMT systems allow remote monitoring of patients with chronic diseases. Thus, it can provide timely patients' diagnostic that can save their life in case of emergencies. However, security in these critical systems is a major challenge facing their wide utilization. In this paper, we present state-of-the-art techniques to secure IoMT systems' data during collection, transmission, and storage. We comprehensively overview IoMT systems' potential attacks, including physical and network attacks. Our findings reveal that most security techniques do not consider various types of attacks. Hence, we propose a security framework that combines several security techniques. The framework covers IoMT security requirements and can mitigate most of its known attacks.

I. INTRODUCTION

IoMT systems are expanding healthcare applications but face serious security challenges because sensitive patient data must be protected across collection, transmission, and storage. The paper reviews existing techniques and attacks, argues that no single technique is comprehensive, and proposes a combined security framework.

  • Motivation: IoMT is an IoT branch for healthcare, expected to represent 40% of the IoT market by the end of 2020.The paper also reports potential healthcare savings of up to $300 billion through greater use of IoMT devices.
  • Motivation: IoMT security is a major challenge because nearly half of devices are reportedly vulnerable to exploits and healthcare data can affect patients’ lives and privacy.Healthcare data is reported to cost 50 times more than credit-card information on the black market.
  • Paper Scope: The paper reviews security requirements and techniques for protecting IoMT data during collection, transmission, and storage.It also examines the resilience of available techniques against different attack types.
  • Findings: The authors argue that no single security technique can provide comprehensive protection against most known IoMT attacks.The review considers cryptographic and non-cryptographic techniques and their resilience against varied attacks.
  • Contribution: The paper proposes an IoMT security framework that combines features of multiple techniques while accounting for device constraints.The framework is intended to cover security during data collection, transmission, and storage.

II. BACKGROUND

The background distinguishes implantable and wearable IoMT devices and describes an architecture that moves biometric data from sensors through gateways to cloud-based storage and visualization.

  • IoMT Types: IoMT systems comprise implantable medical devices and Internet of Wearable Devices, serving necessary or improved assistance for medical conditions.Examples include pacemakers as implantable devices and smartwatches as wearable devices.
  • IoMT Types: Pacemakers are implantable medical devices that help control abnormal heart rhythms.Wireless IMDs address problems associated with wired devices, while long battery life and low power consumption remain essential.
  • IoMT Types: Wearable devices monitor biometrics such as heart rate and movement, including smartwatches, fall-detection bands, ECG monitors, and blood-pressure monitors.
  • IoMT Systems Architecture: Most IoMT systems use four layers spanning biometric collection through storage and visualization by a physician or patient.The architecture supports communication between implantable devices and gateways.
  • IoMT Systems Architecture: The gateway layer receives unprocessed sensor data and can perform validation, short-term storage, and simple AI-based analysis before sending data to the cloud.Gateways may be smartphones or dedicated access points and are generally more powerful than sensors.

III. IOMT SECURITY MODEL

The IoMT security model organizes protection around data collection, transmission, and storage, with requirements covering confidentiality, integrity, availability, accountability, and access control.

  • IoMT Threats at Different Stages: IoMT security threats are analyzed at data collection, data-in-transit, and data-storage stages across the system’s architecture layers.The paper categorizes countermeasure techniques for these stages and examines them in subsequent sections.
  • IoMT Threats at Different Stages: Data-collection attacks can involve software tampering or hardware manipulation and may threaten patients’ lives by affecting sensor operation.
  • IoMT Threats at Different Stages: Attacks during transmission can manipulate or block sensor data exchanged among the four IoMT architecture layers.
  • IoMT Security Requirements: The model defines 11 security requirements derived from CIANA considerations, including confidentiality, integrity, availability, non-repudiation, and authentication.These requirements apply across data collection, transmission, and storage.
  • IoMT Security Requirements: Authorization restricts authenticated users to permitted commands, while anonymity hides patient and physician identities from unauthorized users.Encryption, access-control techniques, and smart cards are cited as ways to support these requirements.
  • IoMT Security Requirements: Forward/backward secrecy protects future keys after old-key compromise and old keys after current-key compromise.Time-based authentication parameters, such as synchronized time-based keys, can provide this property.

C. IoMT Systems Security Techniques

IoMT security techniques are organized into symmetric, asymmetric, and keyless categories, with cryptographic methods supporting authentication and secure access. Symmetric methods can also provide hierarchical data access and serve as a first factor in two-factor authentication.

  • Technique categories: IoMT security techniques divide into symmetric, asymmetric, and keyless categories.Symmetric and asymmetric techniques use cryptographic algorithms, whereas keyless techniques are non-cryptographic.
  • Symmetric cryptography: Symmetric cryptography uses a shared secret key between communicating nodes, generated and distributed before communication.
  • Symmetric cryptography: Symmetric algorithms support hierarchical patient-data access, secure connections without prior setup, and first-factor authentication paired with biometrics or patterns.

B. Wireless Signal Characteristics

Wireless and behavioral characteristics support key generation and authentication in IoMT systems. The reviewed techniques use RSS variation, hashed parameters, gait patterns, facial recognition, and patient-generated tab patterns for security.

  • Wireless signal characteristics: RSS variation inside versus outside the body can generate keys for implantable medical devices without prior connections.The method exploits randomness in received signal strength as a wireless characteristic.
  • Parameter-based key generation: Hashed sensor identifiers and shared keys can support authenticated key-agreement protocols between sensors and gateways.The approach combines a cryptographic hash function, a symmetric key, and XOR.
  • Behavioral key generation: A gait-based system generates symmetric keys from body-worn IoMT sensors within 10 gait-cycles.The authors claim three times as many bits per gait cycle as similar state-of-the-art techniques.
  • Two-factor authentication: Facial recognition can act as a second factor after shared-key authentication for continuous role-based access.
  • Two-factor authentication: A random patient-performed tab pattern can serve as a second factor before sensitive commands execute.The technique can also keep sensor communication turned off until authentication succeeds.

V. ASYMMETRIC-KEY ALGORITHMS

Asymmetric cryptography uses public and private keys and is integrated into IoMT systems through ECC, hash functions, homomorphic encryption, and digital-signature mechanisms. The reviewed approaches target lightweight security, confidential cloud processing, and reduced overhead.

  • Asymmetric cryptography: Asymmetric cryptography uses public and private keys for encryption, decryption, validation, or signatures.RSA and ECC are examples, and ECC is described as common in IoMT because of its lightweight characteristics.
  • Asymmetric-key integration: Asymmetric keys can integrate with hash functions, homomorphic encryption, digital signatures, and two-factor authentication.
  • CHF with ECC: Combining ECC keys with a cryptographic hash function supports a certificateless channel and secure key sharing across IoMT layers.
  • CHF with ECC: 30% less average energy consumption is reported for an ECC-and-CHF technique than for similar techniques.The approach divides patient data into subsets and securely shares them among system entities.
  • Homomorphic encryption: Homomorphic encryption stores patient data as ciphertext while permitting limited mathematical operations in the cloud.Fully homomorphic encryption supports unlimited operations and can enable confidential data aggregation.
  • Homomorphic encryption: 2% to 8% greater breaking time is reported for Optimal HE than for other HE and non-HE techniques individually.Optimal HE modifies fully homomorphic encryption using Step-size Firefly Optimization and also reduces computation time.

C. Digital Signatures

Digital signatures verify data or command authenticity, while keyless and biometric techniques provide alternative or additional authentication mechanisms. Fingerprint biometrics reduce transmission and computation costs relative to ECG-based techniques, with extraction algorithms affecting performance.

  • Digital signatures: Digital signatures verify data or command authenticity using the sender's private key for signing and public key for verification.They can be integrated into sensor firmware through an add-on software shim that validates wireless communications.
  • Multi-factor authentication: Physical keys can provide a second authentication factor alongside ECC keys and improve resistance when one factor is stolen or lost.
  • Keyless techniques: Keyless IoMT security techniques include biometric, token-based, proxy-based, blockchain, and AI approaches.
  • Biometrics: Fingerprint sensors reduce message size and computational overhead during transmission compared with ECG-based techniques.
  • Biometrics: Delaunay Triangulation-based fingerprint extraction shows better performance and lower complexity than the other listed extraction techniques.The cited comparison includes Pair-polar coordinate-based and Minutia Cylinder-Code-based representations.

B. Token-Based Security

Token-based security techniques authenticate users or devices through software and hardware tokens, middleware proxies, or light-based communication. Blockchain and AI extend protection through shared data management and anomaly detection, but latency and data requirements constrain some deployments.

  • Token-Based Security: Software and hardware tokens secure connections between cloud services, IoMT nodes, and hospital sensors.Examples include HTTP x-auth-token browser fields and RFID hardware tokens.
  • Token-Based Security: Proxy-based systems place middleware between sensors and communicating devices to provide full-duplex secure communications.The middleware may use microprocessors embedded in patient-worn jackets or belts.
  • Token-Based Security: Li-Fi can secure hospital monitoring by avoiding wireless-network interference and limiting operational coverage.The passage also identifies substantial free operating frequency as a benefit.
  • Token-Based Security: Blockchain manages shared IoMT data among patients, doctors, and insurers, while AI detects anomalous behavior in network flows and patient data.These approaches are mainly adopted in the cloud layer.
  • Token-Based Security: Public blockchain technology can introduce high latency, and AI-based rare-attack detection requires large amounts of data.Private blockchains may therefore be considered for real-time systems.
  • Token-Based Security: Across the summarized attack surface, 11 of 14 countermeasures use keyless methods and more than half use two-factor authentication.The passage attributes their popularity to implementation and management simplicity.

A. Physical Attacks

IoMT physical and communication attacks target accessible components, transmitted information, authentication factors, or availability. Countermeasures include multifactor authentication, keyless cryptography, DTLS, blockchain, and AI-based mitigation.

  • A. Physical Attacks: Physical attacks require attacker access to IoMT components such as sensors or physical keys to extract patient data or security keys.They target the physical components of the system.
  • A. Physical Attacks: Stealing a smart card can violate authentication, authorization, anonymity, and forward secrecy, while ECC combined with smart cards makes the stolen card insufficient for hijacking.The cited approach uses asymmetric ECC keys with the smart card.
  • A. Physical Attacks: Side-channel attacks leak information during cryptographic communications and can be alleviated with keyless cryptography or DTLS.Blockchain and AI are also identified as detection and mitigation strategies.
  • A. Physical Attacks: Communication-layer attacks can steal or fabricate patient data or block links between IoMT layers, including Bluetooth and Internet connections.The affected links are presented as inter-layer communications.
  • A. Physical Attacks: DoS and DDoS attacks overload communication links, can fragment networks, and commonly target the cloud layer to disrupt availability.Blockchain and AI may reduce effects by finding alternative routes or terminating attacker connections.
  • A. Physical Attacks: Sniffing passively intercepts transmissions between nodes and violates patient-data confidentiality.In IoMT, attackers may observe data sent between architectural layers.
  • A. Physical Attacks: Brute-force and dictionary attacks exploit weak or default passwords, violating authentication and authorization; biometrics are cited as a keyless mitigation.Captured encrypted data may also be attacked offline using powerful machines.

VIII. PROPOSED SECURITY FRAMEWORK FOR IOMT

The proposed framework combines multiple techniques across IoMT security stages because no single technique secures the system against all known attacks. It combines authentication, access control, proxy protection, edge computing, and monitoring to address security requirements and attacks.

  • VIII. PROPOSED SECURITY FRAMEWORK FOR IOMT: The framework is designed to protect IoMT systems from 14 attacks while fulfilling their security requirements.It is organized into three parts based on the IoMT security-model stages.
  • A. Securing Data Collection: Data collection is protected through two-factor authentication, ECC first factors, hierarchical access control, biometrics, and proxy-based techniques.These techniques secure interactions with IoMT systems and protect the patient-data collection stage.
  • A. Securing Data Collection: If one authentication factor is compromised, the other can still provide essential overall security.The framework uses ECC keys as a lightweight first factor and biometrics as a convenient second factor.
  • A. Securing Data Collection: Hardware attacks require an alerting mechanism when the gateway cannot connect to an IoMT sensor for a specified period.The passage gives one hour as an example threshold for alerting the patient and physician.
  • A. Securing Data Collection: Edge computing in the gateway layer reduces latency, supplies processing resources, and can support AI-based initial analysis of sensor readings.The analysis is intended to address data confidentiality and integrity requirements.
  • A. Securing Data Collection: The proposed techniques provide confidentiality, integrity, authentication, authorization, anonymity, secrecy, key-escrow resilience, and session-key agreement.The framework is described as resilient to physical-token, impersonation, tampering, side-channel, sniffing, MITM, relay, replay, synchronization, parallel-session, and brute-force attacks.
  • A. Securing Data Collection: The authentication and access-control techniques assume pre-shared keys or initial parameters, creating setup and emergency-access challenges.The framework requires initial manual KGS preparation, and access may become unusable if the second factor is unavailable.

B. Securing Data in Transit

The framework secures IoMT data in transit with protocols suited to constrained devices and adds AI-based detection for availability and integrity attacks. Its broader deployment still requires manual initial setup and future alternatives for lost authentication factors.

  • B. Securing Data in Transit: CoAP is proposed for communications between sensor and gateway layers, while HTTPS or TLS 1.3 can link the remaining layers.CoAP is designed for resource-constrained IoT applications such as IoMT systems.
  • B. Securing Data in Transit: The transit-security approach supports repudiation, authentication, and secure key exchange but requires initial manual setup.The setup requirement is described as similar to that of the data-collection techniques.
  • B. Securing Data in Transit: AI techniques can detect DoS/DDoS, RF jamming, and stepping-stone attacks by combining detection models with mitigation actions.DNNs are given as an example for intrusion-detection models, followed by termination of compromised connections.
  • IX. CONCLUSIONS AND FUTURE WORK: The proposed framework combines techniques because no single technique satisfies all IoMT security requirements or mitigates most attacks.It covers data and device security from collection through storage and sharing.
  • IX. CONCLUSIONS AND FUTURE WORK: Future work should support a remotely secure initial setup and alternative access when an authentication factor is lost or during emergencies.These needs are identified as challenges of the proposed framework.

Author Biographies:

The author biographies describe an interdisciplinary team with backgrounds in computer engineering, networking, cybersecurity, and IoT-related research across Washington University and Qatar University.

  • Ali Ghubaish: Ali Ghubaish is pursuing a Ph.D. in computer engineering at Washington University in St. Louis after earning computer engineering degrees and working as a teaching assistant.His background includes a networking minor and prior teaching experience at Prince Sattam Bin Abdulaziz University.
  • Tara Salman: Tara Salman is pursuing a Ph.D. in computer engineering at Washington University in St. Louis and previously researched physical-layer security at Qatar University.She earned degrees in computer engineering and computing, including a networking minor.
  • Maede Zolanvari: Maede Zolanvari is a Ph.D. candidate whose research interests include the Internet of Things, machine learning, cybersecurity, secure computer networks, and wireless communications.Her earlier research focused on improving communication-network performance, including OFDM systems.
  • Devrim Unal: Devrim Unal is a Research Assistant Professor of Cyber Security at Qatar University whose interests include cyber-physical systems, IoT security, wireless security, and artificial intelligence.His academic training includes telematics and computer engineering.
  • Abdulla Al-Ali: Abdulla Al-Ali leads Qatar University’s Computer Science and Engineering department and researches cognitive radios for smart cities and vehicular ad-hoc networks.He holds graduate degrees in software design engineering and computer engineering.
  • Raj Jain: Raj Jain is a computer science and engineering professor at Washington University in St. Louis and a Fellow of the ACM.His degrees span electrical engineering, automation, and applied mathematics with computer science.
Loading 2302.04439v1…